Puppet Monit Module

Module for configuring Monit and and setting up monitoring of running processes.

Tested on Debian GNU/Linux 6.0 Squeeze. Patches for other operating systems welcome.

The original module has been extended to support OpenStack implementations. Please read the Usage Section so you understand the details of deploying Monit for OpenStack. The Monit for OpenStack additionas have been tested on Ubuntu Precise and OpenStack Essex.

Installation

Clone this repo to a monit directory under your Puppet modules directory:

git clone git://github.com/danehans/puppet-module-monit.git monit

If you don't have a Puppet Master you can create a manifest file based on the notes below and run Puppet in stand-alone mode providing the module directory you cloned this repo to:

puppet apply --modulepath=modules test_monit.pp

Usage

To install and configure Monit, include the module:

include monit

You can override defaults in the Monit config by including the module with this special syntax:

class { monit: admin => "eivind@uggedal.com", interval => 30 }

Setting up monitoring of processes is done with the monit::site resource. Note that the name needs to be the same as an init script in /etc/init.d:

monit::monitor { "ssh":
  pidfile => "/var/run/sshd.pid",
}

Here is a list of supported OpenStack service definitions. Make sure you only define services that are expected to run on the OpenStack node. Replace $controller_mgt_ip with the IP address or variable of your OpenStack node.

monit::monitor { 'ntp':}

monit::monitor { 'ssh':}

monit::monitor { 'apache2':}

monit::monitor { 'memcached':}

monit::monitor { 'rabbitmq-server':}

monit::monitor { 'novnc':}

monit::monitor { 'keystone':
  monitor_address => $controller_mgt_ip,
}

monit::monitor { 'glance-api':
  monitor_address => $controller_mgt_ip,
}

monit::monitor { 'glance-registry':
  monitor_address => $controller_mgt_ip,
}

monit::monitor { 'nova-api':
  monitor_address => $controller_mgt_ip,
}

You can specify a IP port to check if you're running a network process:

monit::monitor { "nginx":
  pidfile => "/var/run/nginx.pid",
  ip_port => 22,
}

Alternatively you can specify a Unix socket to check:

monit::monitor { "gunicorn-blog":
  pidfile => "/var/run/gunicorn/blog.pid",
  socket => "/var/run/gunicorn/blog.sock",
}

You can also provide additional checks:

$reload_blog = "/etc/init.d/gunicorn-blog reload"

monit::monitor { "gunicorn-blog":
  pidfile => "/var/run/gunicorn/blog.pid",
  socket => "/var/run/gunicorn/blog.sock",
  checks => ["if totalmem > 300 MB for 2 cycles then exec \"$reload_blog\"",
             "if totalmem > 300 MB for 3 cycles then restart",
             "if cpu > 50% for 2 cycles then alert"],
}